how to clean the volume files when volume server down

[ZStephen]

Weed Version : 2.64

My seaweedfs volume server are managed by systemd. When the volume server process's open files reach the limit, the volume server print out lots of logs like this:

When the application try to upload files, The volume server DO NOT stop to assign new volume but still to try to create new ones .So a lots of volume data files such as .ldb and .vif files created in the data directory, more than 1200000. And the volumeId reached 999999 at last.

I try to add the following config to systemd service file and try to reboot the volume server ,it start failed because of OOM.
LimitCORE=infinity
LimitNOFILE=2024000
LimitNPROC=2024000

I can see that the volume server try to load all the volumes, then the kernel killed it because of OOM. 

I know that I can clean the empty volume using weed shell.But now I can NOT start the volume server.   Is any one can help or give some advice? 

Thanks in advanced.

[Chris Lu]

You can directly ”rm” these volumes.

--
You received this message because you are subscribed to the Google Groups "Seaweed File System" group.
To unsubscribe from this group and stop receiving emails from it, send an email to seaweedfs+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/seaweedfs/8db02118-dcb1-4eac-9f41-84cc7bea3b5bn%40googlegroups.com.

[ZStephen]

I write a shell to get all the empty volumes' files and "rm" them all. At last, I started the volume server successful.

Is any plan to optimize weed ?  avoid to create millions of empty volume.

[Chris Lu]

PR is welcome!

To view this discussion on the web visit https://groups.google.com/d/msgid/seaweedfs/a810988f-b864-4c5f-809a-92bb82cf9e82n%40googlegroups.com.